A method and system for identifying hourly-scale temperature rapid transition events
By acquiring hourly temperature data for preprocessing and feature extraction, the system identifies abrupt temperature changes, solving the problem of difficulty in identifying rapid temperature shifts in existing technologies and achieving high temporal resolution temperature monitoring and early warning capabilities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2026-05-11
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies struggle to identify rapid temperature shifts between warm and cold states within a short period, and existing methods have low temporal resolution, failing to meet the needs of power dispatching, human thermal stress monitoring, and refined disaster early warning.
By acquiring hourly temperature data for preprocessing, the magnitude and duration of temperature changes during cold-to-warm and warm-to-cold events are calculated. A moving percentile relative threshold is constructed, a discriminant function for abrupt temperature changes is defined, and event identification is performed in conjunction with the duration of temperature changes. Deduplication and multi-dimensional feature extraction are then carried out.
It enables precise identification of hourly temperature abrupt changes, improving identification accuracy and stability. It can quantitatively characterize the evolution process and severity of events, providing a scientific reference for meteorological disaster early warning and risk assessment.
Smart Images

Figure CN122451702A_ABST